VOICE SAMPLES

Tap play. Every file on this page is a real file served from this site.
3 PLAYABLE SAMPLES
am_michael
Kokoro-82M — local neural TTS, male / US. Runs on the Mac, no API cost.
BENNY VOICE — THE GOOD ONE
af_heart
Kokoro-82M — local neural TTS, female / US. Current default in voice_config.json.
DEFAULT
BENNY B1 — full read
Kokoro-82M, am_michael voice. Long-form narration, "what benny is", ~72s.
BENNY VOICE — FULL LENGTH
EXCLUDED ON PURPOSE
k27tts.py advertises 7 voices. Only 2 exist on disk. These 5 were listed in the script menu but have no voice file and no sample anywhere on the machine, so they are not on this page — picking one would produce silence: Each would need a separate download from the Kokoro-82M repo before it works.
Audio served as assets alongside this page — no local paths, nothing to break on your phone.
Supersedes OLD-20261015·voices.html (the localhost-only version, left in place).
Verified on disk 2026-10-15.